Hemodynamic data were obtained in 13 cirrhotic patients with severe portal hypertension, undergoing combined hepatic vein, umbilicoportal vein, and superior mesenteric artery catheterization. The relative clearance of indocyanine green, the portohepatic gradient (difference between the free portal venous pressure and the free hepatic venous pressure), and the estimated hepatic blood flow were measured. The portal fraction (PF) of total hepatic blood flow was calculated in all patients using indicator dilution curves obtained from the portal bifurcation, a right hepatic vein, and when possible a left hepatic vein (six cases) after injection of (51)Cr-labeled red blood cells ((51)Cr RBC) into the superior mesenteric artery. Flows were overestimated because of loss of indicator through spontaneous portosystemic shunts; however, the ratio between hepatic and portal indicator dilution curves can be used to calculate the portal fraction of total hepatic blood flow since no extrahepatic shunts existed after the bifurcation of the portal vein (as shown on portography). In 10 patients, 15 series of curves were calculable and the PF varied between 30.1 and 100% (mean ± SE: 71.1 ± 6.2%). In the three other patients, only delayed activity from recirculation was detected from portal and hepatic vein samples and PF was 0%; in these three cases, portography and arteriography revealed spontaneous portacaval shunting with reverse and/or stagnant circulation in the portal vein. In the 13 patients, no correlation existed between PF and the relative clearance of indocyanine green or the portohepatic gradient, parameters generally used as indices of severity in cirrhosis. In 10 patients, no correlation was found between PF and the estimated hepatic blood flow.These data indicate that (51)Cr RBC dilution curves can be used for the estimation of the portal fraction of total hepatic blood flow in conscious cirrhotic patients before portacaval shunts. Using this methodology, it could be assessed whether any critical level of portal fraction exists above which poor clinical results occur after portacaval shunting. This measurement could eventually be helpful in determining the appropriate surgical procedure to be applied in individual cases.